Plan requirement
| Subscription | Suite Growth or higher |
| Also required | Copilot |
| Access | Admin |
Switching Copilot on and deciding which agents get it. The rollout matters more than the settings do.
Turn it on
- Open the AI settings in Admin Center.
- Enable Copilot.
- Assign it to the agents who should have it.
- Set the communication guidelines so suggestions match your tone.
- Tell those agents what it does and what it does not.
Start with a few agents
Five or six who are willing, for two weeks. They will find the cases where suggestions are poor, and their opinion carries weight with the rest of the team in a way an announcement does not.
Set the guidelines before the rollout
Suggestions in a tone nobody uses get dismissed in the first hour, and first impressions of this feature are durable. Ten minutes on guidelines saves the launch.
Say what it does not do
It does not send anything, does not decide anything, and is not a measure of the agent's work. Say all three explicitly, because all three are assumed otherwise.
Do not measure adoption
Counting how often agents accept suggestions turns a tool into an assessment, and agents will accept suggestions they would otherwise improve. Measure handling time and quality if you must measure anything.
Collect what is wrong
Give the pilot group a place to report bad suggestions. Nearly all of them trace to articles, so that list is a knowledge base work list arriving from the people closest to customers.
Then widen
Once the pilot group would object to losing it, roll it out. That is a better signal than any usage statistic.
Comments
0 comments
Article is closed for comments.